MongoDB calculating score from existing fields and putting it into a new field in the same collection - mongodb

Question

I'm working on Mongodb and I have one collection, let's say Collection1.
I have to calculate a score from existing fields in Collection1, and put the result into a new field Field8 in Collection1.
Collection1 :
db.Collection1.find().pretty().limit(2) {
"_id": ObjectId("5717a5d4578f3f2556f300f2"),
"Field1": "XXXX",
"Field2": 0,
"Field3": 169,
"Field4": 230,
"Field5": "...4.67", // This field refer to days in a week
"Field6": "ZZ",
"Field7": "LO"
}, {
"_id": ObjectId("17a5d4575f300f278f3f2556"),
"Field1": "YYYY",
"Field2": 1,
"Field3": 260,
"Field4": 80,
"Field5": "1.3....", // This field refer to days in a week
"Field6": "YY",
"Field7": "PK"
}
So, I have to do some calculations to my first collection's fields with the following formula, but I don't know how to proceed ? :
Score = C1*C2*C3*C4
C1 = 10 + 0.03*field3
C2 = 1 or 0.03 it depends on field2 if it equals 1 or 0
C3 = 1 or 2 .... or 7, it depends on field5 for example C3 for this document "Field5": "...4.67" should return 3, it means three days per week
C4 = 1 or field4^-0.6 if field2 equals 0 or 1
After calculating this score I should put it in new field Field8 in my Collection1 and get something just like this :
db.Collection1.find().pretty().limit(2) {
"_id": ObjectId("5717a5d4578f3f2556f300f2"),
"Field1": "XXXX",
"Field2": 0,
"Field3": 169,
"Field4": 230,
"Field5": "...4.67", // This field refer to days in a week
"Field6": "ZZ",
"Field7": "LO",
"Field8": Score // My calculated score
}, {
"_id": ObjectId("17a5d4575f300f278f3f2556"),
"Field1": "YYYY",
"Field2": 1,
"Field3": 260,
"Field4": 80,
"Field5": "1.3....", // This field refer to days in a week
"Field6": "YY",
"Field7": "PK",
"Field8": Score // My calculated score
}
How can I achieve the above?

Answer 1

Depending on your application needs, you can use the aggregation framework for calculating the score and use the bulkWrite() to update your collection. Consider the following example which uses the $project pipeline step as leeway for the score calculations with the arithmetic operators.
Since logic for calculating C3 in your question is getting a number from 1 to 7 which equals exactly 7 - number of points (.), the only feasible approach I can think of is to store an extra field that holds this value first before doing the aggregation. So your first step would be to create that extra field and you can go about it using the bulkWrite() as follows:
Step 1: Modify schema to accomodate extra daysInWeek field
var counter = 0, bulkUpdateOps = [];
db.collection1.find({
"Field5": { "$exists": true }
}).forEach(function(doc) {
// calculations for getting the number of points in Field5
var points, daysInWeek;
points = (doc.Field5.match(new RegExp(".", "g")) || []).length;
daysInWeek = 7 - points;
bulkUpdateOps.push({
"updateOne": {
"filter": { "_id": doc._id },
"update": {
"$set": { "daysInWeek": daysInWeek }
}
}
});
counter++;
if (counter % 500 == 0) {
db.collection1.bulkWrite(bulkUpdateOps);
bulkUpdateOps = [];
}
});
if (counter % 500 != 0) { db.collection1.bulkWrite(bulkUpdateOps); }
Ideally the above operation can also accomodate calculating the other constants in your question and therefore creating the Field8 as a result. However I believe computations like this should be done on the client and let MongoDB do what it does best on the server.
Step 2: Use aggregate to add Field8 field
Having created that extra field daysInWeek you can then construct an aggregation pipeline that projects the new variables using a cohort of arithmetic operators to do the computation (again, would recommend doing such computations on the application layer). The final projection will be the product of the computed fields which you can then use the aggregate result cursor to iterate and add Field8 to the collection with each document:
var pipeline = [
{
"$project": {
"C1": {
"$add": [
10,
{ "$multiply": [ "$Field3", 0.03 ] }
]
},
"C2": {
"$cond": [
{ "$eq": [ "$Field2", 1 ] },
1,
0.03
]
},
"C3": "$daysInWeek",
"C4": {
"$cond": [
{ "$eq": [ "$Field2", 1 ] },
{ "$pow": [ "$Field4", -0.6 ] },
1
]
}
}
},
{
"$project": {
"Field8": { "$multiply": [ "$C1", "$C2", "$C3", "$C4" ] }
}
}
],
counter = 0,
bulkUpdateOps = [];
db.collection1.aggregate(pipeline).forEach(function(doc) {
bulkUpdateOps.push({
"updateOne": {
"filter": { "_id": doc._id },
"update": {
"$set": { "Field8": doc.Field8 }
}
}
});
counter++;
if (counter % 500 == 0) {
db.collection1.bulkWrite(bulkUpdateOps);
bulkUpdateOps = [];
}
});
if (counter % 500 != 0) { db.collection1.bulkWrite(bulkUpdateOps); }
For MongoDB >= 2.6 and <= 3.0, use the Bulk Opeartions API where you need to iterate the collection using the cursor's forEach() method, update each document in the collection.
Some of the arithmetic operators from the above aggregation pipeline are not available in MongoDB >= 2.6 and <= 3.0 so you will need to do the computations within the forEach() iteration.
Use the bulk API to reduce server write requests by bundling each update in bulk and sending to the server only once in every 500 documents in the collection for processing:
var bulkUpdateOps = db.collection1.initializeUnorderedBulkOp(),
cursor = db.collection1.find(), // cursor
counter = 0;
cursor.forEach(function(doc) {
// computations
var c1, c2, c3, c4, Field8;
c1 = 10 + (0.03*doc.Field3);
c2 = (doc.Field2 == 1) ? 1: 0.03;
c3 = 7 - (doc.Field5.match(new RegExp(".", "g")) || []).length;
c4 = (doc.Field2 == 1) ? Math.pow(doc.Field, -0.6) : 1;
Field8 = c1*c2*c3*c4;
bulkUpdateOps.find({ "_id": doc._id }).updateOne({
"$set": { "Field8": Field8 }
});
if (counter % 500 == 0) {
bulkUpdateOps.execute();
bulkUpdateOps = db.collection1.initializeUnorderedBulkOp();
}
})
if (counter % 500 != 0) { bulkUpdateOps.execute(); }

Answer 2

just make a function which returns your calculated value and call in your update mongodb query.
like
var cal = function(row){ return row.Field1 + row.Field2 * row.Field3; // use your formula according to your requirements};
var rows = db.collection1.find() // can use your filter;
rows.forEach(function(row){db.collection1.update({"_id":row._id},{$set:{"Field8":cal(row)}})});